Imagine designing RNA therapeutics as easily as writing a prescription—specifying the desired biological effects and receiving optimized sequences in return. This research develops a neural translation framework that transforms therapeutic specifications into RNA designs by learning the complex patterns within human genomic data.

By reimagining RNA design as a translation challenge, the approach leverages architectures inspired by advances in language and protein modeling (BERT and ESM3) to navigate the intricate relationship between sequence composition and therapeutic function. The computational framework aims to accelerate RNA medicine development by generating sequences that achieve targeted therapeutic outcomes while maintaining compatibility with human cellular machinery.

Project Focus Area

• Model Development: Implementing a mask-based deep learning architecture for predicting functional annotations and RNA sequences
• UTR Analysis: Investigating the relationship between coding and non-coding regions for optimized RNA design
• Sequence Optimization: Developing approaches for codon optimization while maintaining functional properties
• Structural Integration: Incorporating secondary and tertiary structural properties into the prediction framework

The Self-Organizing Systems Lab merges practical biological applications of machine learning with strong theoretical foundations. Our interdisciplinary team actively works toward publications, offering students an opportunity to engage with cutting-edge research at the intersection of computational modeling and RNA therapeutics.